Region of Interest extraction is the crucial step in palm vein recognition, which is quite important for the subsequent feature extraction and recognition. In this paper, we propose a novel ROI extraction method which is robust to the effect of scaling, rotation and irregular boundary. It constructs a down-sampling method for the boundary approximation and presents a novel concavity analysis model to extract the ROI with high success rate. Moreover, this paper uses the distance extension factors to extend the normal ROI extraction and obtain an information-richer region. Example results on the PolyU palm vein database and CASIA database indicate that the proposed ROI extraction method gives a superior performance as compared to the previously proposed point-based approaches.